Wire-based calibration apparatus for x-ray imaging systems

Abstract

A calibration target for use with a radiographic image detector includes a target body securable to the image detector and a plurality of radiopaque linear markers, e.g. wires, fixed to the target body, wherein access to the image detector by incident radiation is at least partially blocked by the plurality of linear markers. By using the geometric properties of wires and advanced detection techniques, a precise calibration suitable for high-quality, volumetric three-dimensional CT reconstruction from biplanar X-ray images is achieved.

What is claimed is: 
     
         1 .- 25 . (canceled) 
     
     
         26 . A method of calibration of spatial reference volumes comprising:
 A) detecting a position of a first reference marker proximate a subject, the position of the first reference marker defining six degrees of freedom and rotation and translation data relative to a pose of the first reference marker;   B) detecting a position of a second reference marker proximate a calibration target on a radiographic image detector, the calibration target containing calibration markers with predetermined spatial coordinates relative to the calibration target and visible in biplanar radiographic images detected by the radiographic image detector, the positions of the second reference marker defining six degrees of freedom and rotation and translation data relative to a pose of the second reference marker;   C) determining a position of the calibration markers in at least two biplanar radiographic images detected by the radiographic image detector;   D) calibrating intrinsic and extrinsic parameters of the radiographic image detector, the parameters providing poses of independent image projections of each of the at least two biplanar images, and   E) generating a registration transform from the two independent image projections, the registration transform defining a center and orientation of a voxel grid usable for back projection and reconstruction of a three dimensional volume; and   F) reconstructing a three dimensional volume from the registration transform by back projection of the voxel grid.   
     
     
         27 . A method for calibration of a radiographic image detector, the method comprising:
 A) acquiring a biplanar radiographic image of a subject with linear obstructions present within the biplanar image;   B) detecting positions of the linear obstructions within the biplanar image;   C) generating a labeled masks where each of the linear obstructions is assigned a unique label;   D) fitting a polynomial equation to each of the linear obstructions, the polynomial equation encompassing pixels within the linear obstruction;   E) computing crossover points of the polynomial equations in different planes plane;   F) generating of equally spaced points along the fitted polynomials; and   G) removing the linear obstructions from the biplanar image.   
     
     
         28 . A method for calibration of a radiographic image detector, the method comprising:
 A) acquiring a biplanar radiographic image of a subject with linear obstructions present within the biplanar image;   B) detecting the linear obstructions within the biplanar image with a pretrained deep learning model;   C) generating, with the deep learning model, a labeled masks where each of the linear obstructions is assigned a unique label;   D) fitting a polynomial equation to each of the linear obstructions, the polynomial equation identifying pixels within each of the respective linear obstructions;   E) computing crossover points of the polynomial equations;   F) generation of equally spaced points along the linear obstructions; and   G) removing the linear obstructions from the biplanar image.   
     
     
         29 . A method for calibration of a radiographic image detector, the method comprising:
 A) acquiring a biplanar radiographic image of a subject with linear obstructions present within the biplanar image;   B) segmenting the linear obstructions from the background;   C) assigning unique labels to each of the linear obstructions;   D) removing the linear obstructions from the biplanar image.